小罗碎碎念
2024年即将过去，跟着小罗同学一起来从宏观层面简单分析一下整个领域目前的一个进展吧。
医学AI的应用场景大致可以划分为两类——癌种+场景，我们详细了解一下不同领域的应用情况，对于我们选择课题是极为有帮助的，尤其是新一轮的国自然标书写作大幕已经拉开，精准的把控方向是非常有必要的！
经过在pubmed粗略检索，得到了下面一些有趣的信息，和大家一起分享。首先看一下AI在15个常见癌种方面的应用情况，数量最多的是乳腺癌（6347项），数量最少的是子宫内膜癌（200项）。
再来看一下AI在医学整体领域的29个应用场景，会发现图像分类与识别是涉及的最多的（67349项），生成模型是涉及的最少的（784项）。
我们需要从关注量最大的一批研究中学经验，从关注量不够的领域中找机会，所以我会从癌种调研和场景调研中挑选出四个话题，和大家做一个详细的分析——分别是数量最多的两个和数量最少的两个。
推文内容有限，并且我规定自己每日推文写作时间必须控制在1.5h以内，所以本期推文只是起到一个抛砖引玉的作用，希望大家能有所启发！如果需要推文中提及的表格，欢迎订阅我的知识星球！
一、癌种调研
在15种肿瘤中，与医学AI结合最密切的是乳腺癌（6347项），涉及最少的是子宫内膜癌（200）。
乳腺癌
剔除掉部分低质文献后，乳腺癌相关的研究还有3650项。
先来看一下近年的论文发表情况。
| 年份 | 数量 |
|---|---|
| 2019 | 3 |
| 2020 | 421 |
| 2021 | 522 |
| 2022 | 719 |
| 2023 | 818 |
| 2024 | 1137 |
| 2025 | 29 |
再看一下近年影响因子最高的10篇文章。
| Title | Journal/Book | 最新IF | Create Date | DOI |
|---|---|---|---|---|
| Radiotherapy to regional nodes in early breast cancer: an individual patient data meta-analysis of 14 324 women in 16 trials | Lancet | 98.4 | 2023/11/6 | 10.1016/S0140-6736(23)01082-6 |
| The promise of AI in personalized breast cancer screening: are we there yet? | Nat Rev Clin Oncol | 81.1 | 2024/3/13 | 10.1038/s41571-024-00877-z |
| Towards equitable AI in oncology | Nat Rev Clin Oncol | 81.1 | 2024/6/7 | 10.1038/s41571-024-00909-8 |
| Artificial Intelligence Improves Breast Cancer Screening in Study | JAMA | 63.1 | 2020/2/12 | 10.1001/jama.2020.0370 |
| Robust breast cancer detection in mammography and digital breast tomosynthesis using an annotation-efficient deep learning approach | Nat Med | 58.7 | 2021/1/12 | 10.1038/s41591-020-01174-9 |
| A population-level digital histologic biomarker for enhanced prognosis of invasive breast cancer | Nat Med | 58.7 | 2023/11/27 | 10.1038/s41591-023-02643-7 |
| Machine learning improves prediction of clinical outcomes for invasive breast cancers | Nat Med | 58.7 | 2023/11/30 | 10.1038/s41591-023-02667-z |
| AI as a new paradigm for risk-based screening for breast cancer | Nat Med | 58.7 | 2022/1/14 | 10.1038/s41591-021-01649-3 |
| Setting guidelines to report the use of AI in clinical trials | Nat Med | 58.7 | 2020/9/10 | 10.1038/s41591-020-1069-z |
| Optimizing risk-based breast cancer screening policies with reinforcement learning | Nat Med | 58.7 | 2022/1/14 | 10.1038/s41591-021-01599-w |
其余的深层次信息，考虑到时间原因，小罗不再进一步分析，感兴趣的可以前往知识星球获取表格，自行分析。
子宫内膜癌
剔除掉部分低质量文献，子宫内膜癌相关的研究还有134项。
近年的论文发表情况如下。
| 年份 | 数量 |
|---|---|
| 2020 | 14 |
| 2021 | 22 |
| 2022 | 21 |
| 2023 | 27 |
| 2024 | 47 |
| 2025 | 3 |
再看一下近年影响因子最高的10篇文章。
| Title | Journal/Book | 最新IF | Create Date | DOI |
|---|---|---|---|---|
| Prediction of recurrence risk in endometrial cancer with multimodal deep learning | Nat Med | 58.7 | 2024/5/24 | 10.1038/s41591-024-02993-w |
| Proteogenomic insights suggest druggable pathways in endometrial carcinoma | Cancer Cell | 48.8 | 2023/8/11 | 10.1016/j.ccell.2023.07.007 |
| Prognostic impact and causality of age on oncological outcomes in women with endometrial cancer: a multimethod analysis of the randomised PORTEC-1, PORTEC-2, and PORTEC-3 trials | Lancet Oncol | 41.6 | 2024/5/3 | 10.1016/S1470-2045(24)00142-6 |
| Interpretable deep learning model to predict the molecular classification of endometrial cancer from haematoxylin and eosin-stained whole-slide images: a combined analysis of the PORTEC randomised trials and clinical cohorts | Lancet Digit Health | 23.8 | 2022/12/10 | 10.1016/S2589-7500(22)00210-2 |
| Breakthrough SARS-CoV-2 Infections, Hospitalizations, and Mortality in Vaccinated Patients With Cancer in the US Between December 2020 and November 2021 | JAMA Oncol | 22.5 | 2022/4/8 | 10.1001/jamaoncol.2022.1096 |
| mBodyMap: a curated database for microbes across human body and their associations with health and diseases | Nucleic Acids Res | 16.6 | 2021/10/31 | 10.1093/nar/gkab973 |
| AI-based histopathology image analysis reveals a distinct subset of endometrial cancers | Nat Commun | 14.7 | 2024/6/26 | 10.1038/s41467-024-49017-2 |
| VOLTA: an enVironment-aware cOntrastive ceLl represenTation leArning for histopathology | Nat Commun | 14.7 | 2024/5/10 | 10.1038/s41467-024-48062-1 |
| Artificial intelligence-based risk stratification, accurate diagnosis and treatment prediction in gynecologic oncology | Semin Cancer Biol | 12.1 | 2023/10/2 | 10.1016/j.semcancer.2023.09.005 |
| Development and Validation of Multiparametric MRI-based Radiomics Models for Preoperative Risk Stratification of Endometrial Cancer | Radiology | 12.1 | 2022/7/12 | 10.1148/radiol.212873 |
其余的深层次信息，考虑到时间原因，小罗不再进一步分析，感兴趣的可以前往知识星球获取表格，自行分析。
二、场景调研
在29个不同的应用场景中，研究数量最多的是“图像识别与分类”（67349项），涉及最少的是“生成模型”（784项）。
图像识别与分类
剔除掉部分低质量文献以后，图像识别与分类相关的研究仅剩4316项。
近年的论文发表情况如下。
| 年份 | 数量 |
|---|---|
| 2019 | 4 |
| 2020 | 891 |
| 2021 | 811 |
| 2022 | 834 |
| 2023 | 710 |
| 2024 | 1040 |
| 2025 | 26 |
再看一下近年影响因子最高的10篇文章。
| Title | Journal/Book | 最新IF | Create Date | DOI |
|---|---|---|---|---|
| Context-dependent functions of pattern recognition receptors in cancer | Nat Rev Cancer | 72.5 | 2022/3/31 | 10.1038/s41568-022-00462-5 |
| Damage turns on root PRRs | Nat Rev Immunol | 67.7 | 2020/2/13 | 10.1038/s41577-020-0287-4 |
| Constitutive immune mechanisms: mediators of host defence and immune regulation | Nat Rev Immunol | 67.7 | 2020/8/13 | 10.1038/s41577-020-0391-5 |
| DAMP-sensing receptors in sterile inflammation and inflammatory diseases | Nat Rev Immunol | 67.7 | 2019/9/28 | 10.1038/s41577-019-0215-7 |
| All life senses infections via pattern recognition | Nat Rev Immunol | 67.7 | 2022/9/6 | 10.1038/s41577-022-00780-7 |
| A pathology foundation model for cancer diagnosis and prognosis prediction | Nature | 50.5 | 2024/9/4 | 10.1038/s41586-024-07894-z |
| A peroxisomal ubiquitin ligase complex forms a retrotranslocation channel | Nature | 50.5 | 2022/6/29 | 10.1038/s41586-022-04903-x |
| Structural insights into how Prp5 proofreads the pre-mRNA branch site | Nature | 50.5 | 2021/8/5 | 10.1038/s41586-021-03789-5 |
| Structural insights into inhibitor regulation of the DNA repair protein DNA-PKcs | Nature | 50.5 | 2022/1/6 | 10.1038/s41586-021-04274-9 |
| How your brain detects patterns in the everyday: without conscious thought | Nature | 50.5 | 2024/9/25 | 10.1038/d41586-024-03116-8 |
其余的深层次信息，考虑到时间原因，小罗不再进一步分析，感兴趣的可以前往知识星球获取表格，自行分析。
生成模型
剔除掉部分低分文献后，生成模型相关的研究还有456项。
近年的论文发表情况如下。
| 年份 | 数量 |
|---|---|
| 2020 | 44 |
| 2021 | 47 |
| 2022 | 66 |
| 2023 | 99 |
| 2024 | 200 |
再看一下近年影响因子最高的10篇文章。
| Title | Journal/Book | 最新IF | Create Date | DOI |
|---|---|---|---|---|
| AI-Generated Medical Advice-GPT and Beyond | JAMA | 63.1 | 2023/3/27 | 10.1001/jama.2023.5321 |
| AI-Generated Clinical Summaries | JAMA | 63.1 | 2024/5/16 | 10.1001/jama.2024.7277 |
| AI-Generated Clinical Summaries-Reply | JAMA | 63.1 | 2024/5/16 | 10.1001/jama.2024.7280 |
| AI-Generated Clinical Summaries Require More Than Accuracy | JAMA | 63.1 | 2024/1/29 | 10.1001/jama.2024.0555 |
| Caution with AI-generated content in biomedicine | Nat Med | 58.7 | 2023/2/7 | 10.1038/d41591-023-00014-w |
| AI-generated text may have a role in evidence-based medicine | Nat Med | 58.7 | 2023/5/23 | 10.1038/s41591-023-02366-9 |
| Avoid patenting AI-generated inventions | Nature | 50.5 | 2023/10/3 | 10.1038/d41586-023-03116-0 |
| ‘ChatGPT detector’ catches AI-generated papers with unprecedented accuracy | Nature | 50.5 | 2023/11/17 | 10.1038/d41586-023-03479-4 |
| Don’t let watermarks stigmatize AI-generated research content | Nature | 50.5 | 2024/11/26 | 10.1038/d41586-024-03869-2 |
| AI-generated images and video are here: how could they shape research? | Nature | 50.5 | 2024/3/7 | 10.1038/d41586-024-00659-8 |
| AI produces gibberish when trained on too much AI-generated data | Nature | 50.5 | 2024/7/24 | 10.1038/d41586-024-02355-z |
其余的深层次信息，考虑到时间原因，小罗不再进一步分析，感兴趣的可以前往知识星球获取表格，自行分析。
知识星球
如需获取推文中提及的各种资料，欢迎加入我的知识星球！
扫一扫
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
